poll_mode_drv.rst revision 97f17497
1..  BSD LICENSE
2    Copyright(c) 2010-2015 Intel Corporation. All rights reserved.
3    All rights reserved.
4
5    Redistribution and use in source and binary forms, with or without
6    modification, are permitted provided that the following conditions
7    are met:
8
9    * Redistributions of source code must retain the above copyright
10    notice, this list of conditions and the following disclaimer.
11    * Redistributions in binary form must reproduce the above copyright
12    notice, this list of conditions and the following disclaimer in
13    the documentation and/or other materials provided with the
14    distribution.
15    * Neither the name of Intel Corporation nor the names of its
16    contributors may be used to endorse or promote products derived
17    from this software without specific prior written permission.
18
19    THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
20    "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
21    LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
22    A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
23    OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
24    SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
25    LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
26    DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
27    THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
28    (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
29    OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
30
31.. _Poll_Mode_Driver:
32
33Poll Mode Driver
34================
35
36The DPDK includes 1 Gigabit, 10 Gigabit and 40 Gigabit and para virtualized virtio Poll Mode Drivers.
37
38A Poll Mode Driver (PMD) consists of APIs, provided through the BSD driver running in user space,
39to configure the devices and their respective queues.
40In addition, a PMD accesses the RX and TX descriptors directly without any interrupts
41(with the exception of Link Status Change interrupts) to quickly receive,
42process and deliver packets in the user's application.
43This section describes the requirements of the PMDs,
44their global design principles and proposes a high-level architecture and a generic external API for the Ethernet PMDs.
45
46Requirements and Assumptions
47----------------------------
48
49The DPDK environment for packet processing applications allows for two models, run-to-completion and pipe-line:
50
51*   In the *run-to-completion*  model, a specific port's RX descriptor ring is polled for packets through an API.
52    Packets are then processed on the same core and placed on a port's TX descriptor ring through an API for transmission.
53
54*   In the *pipe-line*  model, one core polls one or more port's RX descriptor ring through an API.
55    Packets are received and passed to another core via a ring.
56    The other core continues to process the packet which then may be placed on a port's TX descriptor ring through an API for transmission.
57
58In a synchronous run-to-completion model,
59each logical core assigned to the DPDK executes a packet processing loop that includes the following steps:
60
61*   Retrieve input packets through the PMD receive API
62
63*   Process each received packet one at a time, up to its forwarding
64
65*   Send pending output packets through the PMD transmit API
66
67Conversely, in an asynchronous pipe-line model, some logical cores may be dedicated to the retrieval of received packets and
68other logical cores to the processing of previously received packets.
69Received packets are exchanged between logical cores through rings.
70The loop for packet retrieval includes the following steps:
71
72*   Retrieve input packets through the PMD receive API
73
74*   Provide received packets to processing lcores through packet queues
75
76The loop for packet processing includes the following steps:
77
78*   Retrieve the received packet from the packet queue
79
80*   Process the received packet, up to its retransmission if forwarded
81
82To avoid any unnecessary interrupt processing overhead, the execution environment must not use any asynchronous notification mechanisms.
83Whenever needed and appropriate, asynchronous communication should be introduced as much as possible through the use of rings.
84
85Avoiding lock contention is a key issue in a multi-core environment.
86To address this issue, PMDs are designed to work with per-core private resources as much as possible.
87For example, a PMD maintains a separate transmit queue per-core, per-port.
88In the same way, every receive queue of a port is assigned to and polled by a single logical core (lcore).
89
90To comply with Non-Uniform Memory Access (NUMA), memory management is designed to assign to each logical core
91a private buffer pool in local memory to minimize remote memory access.
92The configuration of packet buffer pools should take into account the underlying physical memory architecture in terms of DIMMS,
93channels and ranks.
94The application must ensure that appropriate parameters are given at memory pool creation time.
95See :ref:`Mempool Library <Mempool_Library>`.
96
97Design Principles
98-----------------
99
100The API and architecture of the Ethernet* PMDs are designed with the following guidelines in mind.
101
102PMDs must help global policy-oriented decisions to be enforced at the upper application level.
103Conversely, NIC PMD functions should not impede the benefits expected by upper-level global policies,
104or worse prevent such policies from being applied.
105
106For instance, both the receive and transmit functions of a PMD have a maximum number of packets/descriptors to poll.
107This allows a run-to-completion processing stack to statically fix or
108to dynamically adapt its overall behavior through different global loop policies, such as:
109
110*   Receive, process immediately and transmit packets one at a time in a piecemeal fashion.
111
112*   Receive as many packets as possible, then process all received packets, transmitting them immediately.
113
114*   Receive a given maximum number of packets, process the received packets, accumulate them and finally send all accumulated packets to transmit.
115
116To achieve optimal performance, overall software design choices and pure software optimization techniques must be considered and
117balanced against available low-level hardware-based optimization features (CPU cache properties, bus speed, NIC PCI bandwidth, and so on).
118The case of packet transmission is an example of this software/hardware tradeoff issue when optimizing burst-oriented network packet processing engines.
119In the initial case, the PMD could export only an rte_eth_tx_one function to transmit one packet at a time on a given queue.
120On top of that, one can easily build an rte_eth_tx_burst function that loops invoking the rte_eth_tx_one function to transmit several packets at a time.
121However, an rte_eth_tx_burst function is effectively implemented by the PMD to minimize the driver-level transmit cost per packet through the following optimizations:
122
123*   Share among multiple packets the un-amortized cost of invoking the rte_eth_tx_one function.
124
125*   Enable the rte_eth_tx_burst function to take advantage of burst-oriented hardware features (prefetch data in cache, use of NIC head/tail registers)
126    to minimize the number of CPU cycles per packet, for example by avoiding unnecessary read memory accesses to ring transmit descriptors,
127    or by systematically using arrays of pointers that exactly fit cache line boundaries and sizes.
128
129*   Apply burst-oriented software optimization techniques to remove operations that would otherwise be unavoidable, such as ring index wrap back management.
130
131Burst-oriented functions are also introduced via the API for services that are intensively used by the PMD.
132This applies in particular to buffer allocators used to populate NIC rings, which provide functions to allocate/free several buffers at a time.
133For example, an mbuf_multiple_alloc function returning an array of pointers to rte_mbuf buffers which speeds up the receive poll function of the PMD when
134replenishing multiple descriptors of the receive ring.
135
136Logical Cores, Memory and NIC Queues Relationships
137--------------------------------------------------
138
139The DPDK supports NUMA allowing for better performance when a processor's logical cores and interfaces utilize its local memory.
140Therefore, mbuf allocation associated with local PCIe* interfaces should be allocated from memory pools created in the local memory.
141The buffers should, if possible, remain on the local processor to obtain the best performance results and RX and TX buffer descriptors
142should be populated with mbufs allocated from a mempool allocated from local memory.
143
144The run-to-completion model also performs better if packet or data manipulation is in local memory instead of a remote processors memory.
145This is also true for the pipe-line model provided all logical cores used are located on the same processor.
146
147Multiple logical cores should never share receive or transmit queues for interfaces since this would require global locks and hinder performance.
148
149Device Identification and Configuration
150---------------------------------------
151
152Device Identification
153~~~~~~~~~~~~~~~~~~~~~
154
155Each NIC port is uniquely designated by its (bus/bridge, device, function) PCI
156identifiers assigned by the PCI probing/enumeration function executed at DPDK initialization.
157Based on their PCI identifier, NIC ports are assigned two other identifiers:
158
159*   A port index used to designate the NIC port in all functions exported by the PMD API.
160
161*   A port name used to designate the port in console messages, for administration or debugging purposes.
162    For ease of use, the port name includes the port index.
163
164Device Configuration
165~~~~~~~~~~~~~~~~~~~~
166
167The configuration of each NIC port includes the following operations:
168
169*   Allocate PCI resources
170
171*   Reset the hardware (issue a Global Reset) to a well-known default state
172
173*   Set up the PHY and the link
174
175*   Initialize statistics counters
176
177The PMD API must also export functions to start/stop the all-multicast feature of a port and functions to set/unset the port in promiscuous mode.
178
179Some hardware offload features must be individually configured at port initialization through specific configuration parameters.
180This is the case for the Receive Side Scaling (RSS) and Data Center Bridging (DCB) features for example.
181
182On-the-Fly Configuration
183~~~~~~~~~~~~~~~~~~~~~~~~
184
185All device features that can be started or stopped "on the fly" (that is, without stopping the device) do not require the PMD API to export dedicated functions for this purpose.
186
187All that is required is the mapping address of the device PCI registers to implement the configuration of these features in specific functions outside of the drivers.
188
189For this purpose,
190the PMD API exports a function that provides all the information associated with a device that can be used to set up a given device feature outside of the driver.
191This includes the PCI vendor identifier, the PCI device identifier, the mapping address of the PCI device registers, and the name of the driver.
192
193The main advantage of this approach is that it gives complete freedom on the choice of the API used to configure, to start, and to stop such features.
194
195As an example, refer to the configuration of the IEEE1588 feature for the Intel® 82576 Gigabit Ethernet Controller and
196the Intel® 82599 10 Gigabit Ethernet Controller controllers in the testpmd application.
197
198Other features such as the L3/L4 5-Tuple packet filtering feature of a port can be configured in the same way.
199Ethernet* flow control (pause frame) can be configured on the individual port.
200Refer to the testpmd source code for details.
201Also, L4 (UDP/TCP/ SCTP) checksum offload by the NIC can be enabled for an individual packet as long as the packet mbuf is set up correctly. See `Hardware Offload`_ for details.
202
203Configuration of Transmit and Receive Queues
204~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
205
206Each transmit queue is independently configured with the following information:
207
208*   The number of descriptors of the transmit ring
209
210*   The socket identifier used to identify the appropriate DMA memory zone from which to allocate the transmit ring in NUMA architectures
211
212*   The values of the Prefetch, Host and Write-Back threshold registers of the transmit queue
213
214*   The *minimum* transmit packets to free threshold (tx_free_thresh).
215    When the number of descriptors used to transmit packets exceeds this threshold, the network adaptor should be checked to see if it has written back descriptors.
216    A value of 0 can be passed during the TX queue configuration to indicate the default value should be used.
217    The default value for tx_free_thresh is 32.
218    This ensures that the PMD does not search for completed descriptors until at least 32 have been processed by the NIC for this queue.
219
220*   The *minimum*  RS bit threshold. The minimum number of transmit descriptors to use before setting the Report Status (RS) bit in the transmit descriptor.
221    Note that this parameter may only be valid for Intel 10 GbE network adapters.
222    The RS bit is set on the last descriptor used to transmit a packet if the number of descriptors used since the last RS bit setting,
223    up to the first descriptor used to transmit the packet, exceeds the transmit RS bit threshold (tx_rs_thresh).
224    In short, this parameter controls which transmit descriptors are written back to host memory by the network adapter.
225    A value of 0 can be passed during the TX queue configuration to indicate that the default value should be used.
226    The default value for tx_rs_thresh is 32.
227    This ensures that at least 32 descriptors are used before the network adapter writes back the most recently used descriptor.
228    This saves upstream PCIe* bandwidth resulting from TX descriptor write-backs.
229    It is important to note that the TX Write-back threshold (TX wthresh) should be set to 0 when tx_rs_thresh is greater than 1.
230    Refer to the Intel® 82599 10 Gigabit Ethernet Controller Datasheet for more details.
231
232The following constraints must be satisfied for tx_free_thresh and tx_rs_thresh:
233
234*   tx_rs_thresh must be greater than 0.
235
236*   tx_rs_thresh must be less than the size of the ring minus 2.
237
238*   tx_rs_thresh must be less than or equal to tx_free_thresh.
239
240*   tx_free_thresh must be greater than 0.
241
242*   tx_free_thresh must be less than the size of the ring minus 3.
243
244*   For optimal performance, TX wthresh should be set to 0 when tx_rs_thresh is greater than 1.
245
246One descriptor in the TX ring is used as a sentinel to avoid a hardware race condition, hence the maximum threshold constraints.
247
248.. note::
249
250    When configuring for DCB operation, at port initialization, both the number of transmit queues and the number of receive queues must be set to 128.
251
252Hardware Offload
253~~~~~~~~~~~~~~~~
254
255Depending on driver capabilities advertised by
256``rte_eth_dev_info_get()``, the PMD may support hardware offloading
257feature like checksumming, TCP segmentation or VLAN insertion.
258
259The support of these offload features implies the addition of dedicated
260status bit(s) and value field(s) into the rte_mbuf data structure, along
261with their appropriate handling by the receive/transmit functions
262exported by each PMD. The list of flags and their precise meaning is
263described in the mbuf API documentation and in the in :ref:`Mbuf Library
264<Mbuf_Library>`, section "Meta Information".
265
266Poll Mode Driver API
267--------------------
268
269Generalities
270~~~~~~~~~~~~
271
272By default, all functions exported by a PMD are lock-free functions that are assumed
273not to be invoked in parallel on different logical cores to work on the same target object.
274For instance, a PMD receive function cannot be invoked in parallel on two logical cores to poll the same RX queue of the same port.
275Of course, this function can be invoked in parallel by different logical cores on different RX queues.
276It is the responsibility of the upper-level application to enforce this rule.
277
278If needed, parallel accesses by multiple logical cores to shared queues can be explicitly protected by dedicated inline lock-aware functions
279built on top of their corresponding lock-free functions of the PMD API.
280
281Generic Packet Representation
282~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
283
284A packet is represented by an rte_mbuf structure, which is a generic metadata structure containing all necessary housekeeping information.
285This includes fields and status bits corresponding to offload hardware features, such as checksum computation of IP headers or VLAN tags.
286
287The rte_mbuf data structure includes specific fields to represent, in a generic way, the offload features provided by network controllers.
288For an input packet, most fields of the rte_mbuf structure are filled in by the PMD receive function with the information contained in the receive descriptor.
289Conversely, for output packets, most fields of rte_mbuf structures are used by the PMD transmit function to initialize transmit descriptors.
290
291The mbuf structure is fully described in the :ref:`Mbuf Library <Mbuf_Library>` chapter.
292
293Ethernet Device API
294~~~~~~~~~~~~~~~~~~~
295
296The Ethernet device API exported by the Ethernet PMDs is described in the *DPDK API Reference*.
297
298Extended Statistics API
299~~~~~~~~~~~~~~~~~~~~~~~
300
301The extended statistics API allows each individual PMD to expose a unique set
302of statistics. The client of the API provides an array of
303``struct rte_eth_xstats`` type. Each ``struct rte_eth_xstats`` contains a
304string and value pair. The amount of xstats exposed, and position of the
305statistic in the array must remain constant during runtime.
306
307A naming scheme exists for the strings exposed to clients of the API. This is
308to allow scraping of the API for statistics of interest. The naming scheme uses
309strings split by a single underscore ``_``. The scheme is as follows:
310
311* direction
312* detail 1
313* detail 2
314* detail n
315* unit
316
317Examples of common statistics xstats strings, formatted to comply to the scheme
318proposed above:
319
320* ``rx_bytes``
321* ``rx_crc_errors``
322* ``tx_multicast_packets``
323
324The scheme, although quite simple, allows flexibility in presenting and reading
325information from the statistic strings. The following example illustrates the
326naming scheme:``rx_packets``. In this example, the string is split into two
327components. The first component ``rx`` indicates that the statistic is
328associated with the receive side of the NIC.  The second component ``packets``
329indicates that the unit of measure is packets.
330
331A more complicated example: ``tx_size_128_to_255_packets``. In this example,
332``tx`` indicates transmission, ``size``  is the first detail, ``128`` etc are
333more details, and ``packets`` indicates that this is a packet counter.
334
335Some additions in the metadata scheme are as follows:
336
337* If the first part does not match ``rx`` or ``tx``, the statistic does not
338  have an affinity with either receive of transmit.
339
340* If the first letter of the second part is ``q`` and this ``q`` is followed
341  by a number, this statistic is part of a specific queue.
342
343An example where queue numbers are used is as follows: ``tx_q7_bytes`` which
344indicates this statistic applies to queue number 7, and represents the number
345of transmitted bytes on that queue.
346